Mullard vs. Williamson 60W UL PP amp. Which one?

[SGregory]

I have been searching and havn't seen a thread that directly compares the Mullard Topology vs. a Williamson Topology for a 60W kt88 UL PP amp. If there is one and I am not searching cleverly enough would you be kind enough to post a link. If not, I am interested in opinions.

I have built three different SE tube amps. My favorite so far is the Tubelab SE using 300B. All sound wonderful, but as I work towards my goal in sound system, I need to biamp and would like more power to handle the dynamics of the bass section. I have built an Aleph5, AlephX monoblock 100W and an aleph30, also wonderful but not tubes. Can't afford the power cost in operating the alephX's as they need their own power circuit for each.

I have been following two recent threads, one of a mullard style and another of a modified Williamson design for UL. I am interested in both and may at some point build both, but today I need to choose one.

What amp would be best suited for this application and why. The speakers are 4 ohms nominal (3.5 actual) with an efficiency of 90 db. Peak impedance is around 23ohm at 36hz ported. The Mid/Upper is also 90db.
Thanks for your feedback.

[Eli Duttman]

Dude,

Both topologies are "classic". IMO, the edge goes to Mullard style simply because it's less vulnerable to instability. People overlook the fact that Williamson used creme de la creme O/P "iron" wound by Partridge. It's all too easy to get phase shift oscillation, when Williamson style is executed around "ordinary" O/P trafos.

[SGregory]

I agree that the Williamson must have good iron from what I read. Is the oscillation in the lower freqencies or at the upper frequencies? If it is lower frequencies then the Mullard is clearly the design for this amps application.

[Eli Duttman]

Look at the number of high pass poles in Williamson topology. Darn tootin it's LF instability.

The 6SN7 is a paradigm of linearity. When Williamson style doesn't "barf", it sounds great.

There is a high pass pole in Mullard style that can get overlooked. It's formed by the 1 MOhm resistor that connects the LTP grids together and the cap. that connects the non-inverting triode to ground. However, it's easy enough to get F₃ of that pole very close to zero. For instance, the corner freq. of 1 MOhm and 470 nF. is 0.34 Hz., little chance of trouble there.

[SGregory]

Thanks for the feedback Eli. I roughly see in the williamson schematic your point for the number of high pass poles. At this point I will turn my focus to the Mullard thread that is going on right now. I hope that it goes to completion.

[Merlinb]

The Mullard also provides some interesting possibilities for modding the EF86 input stage, either by converting it to triode, or UL, or even connecting the screen grid to the cathode of the LTP for a different flavour of UL operation.